CN107256513A - Method and device is recommended in a kind of collocation of object - Google Patents
Method and device is recommended in a kind of collocation of object Download PDFInfo
- Publication number
- CN107256513A CN107256513A CN201710465115.1A CN201710465115A CN107256513A CN 107256513 A CN107256513 A CN 107256513A CN 201710465115 A CN201710465115 A CN 201710465115A CN 107256513 A CN107256513 A CN 107256513A
- Authority
- CN
- China
- Prior art keywords
- collocation
- user
- matching
- candidate
- preference
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000012216 screening Methods 0.000 claims description 23
- 238000007781 pre-processing Methods 0.000 claims description 9
- 238000001914 filtration Methods 0.000 abstract description 11
- 239000013065 commercial product Substances 0.000 abstract description 2
- 239000000047 product Substances 0.000 description 37
- OENIXTHWZWFYIV-UHFFFAOYSA-N 2-[4-[2-[5-(cyclopentylmethyl)-1h-imidazol-2-yl]ethyl]phenyl]benzoic acid Chemical compound OC(=O)C1=CC=CC=C1C(C=C1)=CC=C1CCC(N1)=NC=C1CC1CCCC1 OENIXTHWZWFYIV-UHFFFAOYSA-N 0.000 description 19
- WVAKRQOMAINQPU-UHFFFAOYSA-N 2-[4-[2-[5-(2,2-dimethylbutyl)-1h-imidazol-2-yl]ethyl]phenyl]pyridine Chemical compound N1C(CC(C)(C)CC)=CN=C1CCC1=CC=C(C=2N=CC=CC=2)C=C1 WVAKRQOMAINQPU-UHFFFAOYSA-N 0.000 description 8
- 101100111184 Schizosaccharomyces pombe (strain 972 / ATCC 24843) bag101 gene Proteins 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 238000004364 calculation method Methods 0.000 description 7
- 230000000295 complement effect Effects 0.000 description 6
- 230000008569 process Effects 0.000 description 6
- 230000006399 behavior Effects 0.000 description 5
- 238000010586 diagram Methods 0.000 description 5
- 230000009471 action Effects 0.000 description 3
- 239000003086 colorant Substances 0.000 description 3
- 239000000463 material Substances 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 238000005034 decoration Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 241000565661 Neohelix albolabris Species 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- -1 jack 1 Chemical compound 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0631—Item recommendations
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
Landscapes
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Engineering & Computer Science (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Method and device is recommended in the collocation that the present invention discloses a kind of object, this method and device are in acquisition targeted customer wait after the existing object arranged in pairs or groups, based on the user's collocation preference table for pre-establishing and storing, it is determined that having at least one candidate target of incidence relation with the existing object, and utilize predetermined filtering rule, at least one destination object arranged in pairs or groups with the existing object is filtered out from least one described candidate target, finally, recommended and at least one destination object described in existing object collocation to targeted customer with predetermined exhibition method.It can be seen that, the present invention realizes a kind of scheme for the recommendation that can be arranged in pairs or groups to object progress, and for scenes such as cyber recommendations, the collocation formula that can realize cyber using the present invention program is recommended, so as to the commercial product recommending function of effective abundant network shopping mall, and lift Consumer's Experience.
Description
Technical Field
The invention belongs to the technical field of network commodity recommendation based on machine learning, and particularly relates to a collocation recommendation method and device for an object.
Background
Nowadays, online shopping has become a main shopping mode in people's daily life. For online shopping, each large online shopping mall is generally provided with a commodity recommendation function, and when a user browses a certain commodity, the user is provided with more choices by recommending other commodities which the user may be interested in, so that the user consumption is promoted, and the website income is improved.
There are some related technologies for recommending commodities, which generally recommend other commodities similar to or complementary to the current commodity (the commodity browsed or selected by the user) to the user based on commodity similarity or commodity complementarity. For example, patent US20070168357 a1 describes a method of recommending items similar to the picture of the current item based on the similarity of the picture of the selected item and the recommended item, e.g., if the user selects an article of clothing, the system will recommend to the user other clothing having the same/similar profile, style or color as the selected clothing. For another example, the patent US 7437344B 2 describes a method for recommending a product complementary to a current product, which is based on a set of preset complementary rules to implement the recommendation of the product complementary to the current product, such as assuming that the established complementary rules include "complementary lipstick and lip gloss, pink and white compatible", when the user selects a pink lipstick, the system recommends a white lip gloss to the user.
However, in the field of commodity recommendation, the collocation recommendation of commodities, that is, the recommendation of other commodities which can be used in collocation with the current commodity, is often more suitable for the user's needs, and is more likely to attract the user to purchase than general commodity recommendation. In view of this, there is a need in the art to provide a method for implementing a merchandise collocation recommendation.
Disclosure of Invention
In view of this, the present invention aims to provide a method and an apparatus for recommending matching of an object, which are capable of recommending matching of a product in a scenario such as network product recommendation, thereby enriching the product recommendation function of a network mall and improving user experience.
Therefore, the invention discloses the following technical scheme:
a collocation recommendation method for an object comprises the following steps:
obtaining a current object to be collocated of a target user;
determining at least one candidate object having an association relation with the current object based on a user collocation preference table which is formulated and stored in advance; the user collocation preference table comprises preference value information of a plurality of users on a plurality of collocation pairs, the collocation pairs are object collocation pairs consisting of two different types of objects or feature collocation pairs consisting of features of the two different types of objects, and the candidate objects and the current object or the candidate objects and the corresponding features of the current object are in collocation pairs in the user collocation preference table;
selecting at least one target object which is most matched with the current object from the at least one candidate object by utilizing a preset selection rule based on the preference value information corresponding to the matching pair where the candidate object and the current object are located in the user matching preference table;
and recommending the at least one target object to the target user in a preset presentation mode.
Preferably, the determining at least one candidate object having an association relationship with the current object based on a pre-established and stored user collocation preference table includes:
if the collocation pair in the user collocation preference table is an object collocation pair, then:
finding out at least one candidate object matching pair from a user matching preference table, wherein the candidate object matching pair comprises the current object;
determining other objects except the current object from the at least one candidate object collocation pair as candidate objects;
if the matching pair in the user matching preference table is a feature matching pair, then:
finding out at least one candidate feature matching pair from a user matching preference table, wherein the candidate feature matching pair comprises the corresponding feature of the current object;
and determining other features except the features of the current object from the at least one candidate feature matching pair, and determining an object meeting at least one feature of the other features as a candidate object.
In the above method, preferably, the selecting, based on the preference value information corresponding to the matching pair where the candidate object and the current object are located in the user matching preference table, at least one target object most matched with the current object from the at least one candidate object by using a predetermined selection rule includes:
finding out preference value information of each candidate object and the object matching pair where the current object is located or the feature matching pair corresponding to the target user from the user matching preference table, and/or preference value information corresponding to all users;
calculating the collocation degree of each candidate object and the current object based on the obtained preference value information;
and screening out a preset number of candidate objects with the highest matching degree with the current object as target objects.
Preferably, the recommending the at least one target object to the user in the predetermined presentation manner includes:
recommending the at least one target object to a target user in an object list mode; or,
recommending the at least one target object to a target user in an object collocation combination mode, wherein each recommended object collocation combination is a collocation combination of the at least one target object and the current object.
Preferably, the method further includes, before the obtaining of the current object to be collocated of the target user, a preprocessing step: generating a user collocation preference table;
the generating of the user collocation preference table comprises the following steps:
determining each possible object or feature pairing;
acquiring collection information of each user on the object matching combination, and calculating preference values of each user on object matching pairs or feature matching pairs correspondingly contained in the collected object matching combinations based on the collection information of each user on the object matching combination;
predicting the preference value of the matched pair of which the preference value is not calculated in each possible object matched pair or feature matched pair by each user by using a preset algorithm;
and generating a user matching preference table according to the possible object matching pairs or the feature matching pairs and the preference values of the users to the possible object matching pairs or the feature matching pairs.
A collocation recommendation device for an object, comprising:
the device comprises an acquisition unit, a matching unit and a matching unit, wherein the acquisition unit is used for acquiring a current object to be matched of a target user;
the determining unit is used for determining at least one candidate object which has an incidence relation with the current object based on a user collocation preference table which is formulated and stored in advance; the user collocation preference table comprises preference value information of a plurality of users on a plurality of collocation pairs, the collocation pairs are object collocation pairs consisting of two different types of objects or feature collocation pairs consisting of features of the two different types of objects, and the candidate objects and the current object or the candidate objects and the corresponding features of the current object are in collocation pairs in the user collocation preference table;
a screening unit, configured to screen, based on preference value information corresponding to a matching pair where a candidate object and a current object are located in the user matching preference table, at least one target object most matched with the current object from the at least one candidate object by using a predetermined screening rule;
and the collocation recommending unit is used for recommending the at least one target object to the target user in a preset display mode.
The above apparatus, preferably, the determining unit is further configured to:
if the collocation pair in the user collocation preference table is an object collocation pair, then:
finding out at least one candidate object matching pair from a user matching preference table, wherein the candidate object matching pair comprises the current object; determining other objects except the current object from the at least one candidate object collocation pair as candidate objects;
if the matching pair in the user matching preference table is a feature matching pair, then:
finding out at least one candidate feature matching pair from a user matching preference table, wherein the candidate feature matching pair comprises the corresponding feature of the current object; and determining other features except the features of the current object from the at least one candidate feature matching pair, and determining an object meeting at least one feature of the other features as a candidate object.
The above apparatus, preferably, the screening unit is further configured to:
finding out preference value information of each candidate object and the object matching pair where the current object is located or the feature matching pair corresponding to the target user from the user matching preference table, and/or preference value information corresponding to all users; calculating the collocation degree of each candidate object and the current object based on the obtained preference value information; and screening out a preset number of candidate objects with the highest matching degree with the current object as target objects.
The above apparatus, preferably, the collocation recommending unit is further configured to:
recommending the at least one target object to a target user in an object list mode; or recommending the at least one target object to a target user in an object collocation combination mode, wherein each recommended object collocation combination is a collocation combination of the at least one target object and the current object.
The above apparatus, preferably, further comprises:
the preprocessing unit is used for generating a user collocation preference table;
the preprocessing unit generates a user collocation preference table, comprising:
determining each possible object or feature pairing; acquiring collection information of each user on the object matching combination, and calculating preference values of each user on object matching pairs or feature matching pairs correspondingly contained in the collected object matching combinations based on the collection information of each user on the object matching combination; predicting the preference value of the matched pair of which the preference value is not calculated in each possible object matched pair or feature matched pair by each user by using a preset algorithm; and generating a user matching preference table according to the possible object matching pairs or the feature matching pairs and the preference values of the users to the possible object matching pairs or the feature matching pairs.
According to the above scheme, after the current object to be collocated of the target user is obtained, at least one candidate object having an association relation with the current object is determined based on a user collocation preference table which is formulated and stored in advance, at least one target object collocated with the current object is screened from the at least one candidate object by using a preset screening rule, and finally, the at least one target object collocated with the current object is recommended to the target user in a preset display mode. Therefore, the invention realizes a scheme capable of matching and recommending the object, and can realize matching and recommending of the network commodities by utilizing the scheme of the invention aiming at scenes such as network commodity recommendation and the like, thereby effectively enriching the commodity recommendation function of the network shopping mall and improving the user experience.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a flowchart of a method for recommending matching of an object according to an embodiment of the present invention;
FIG. 2 is an example of a merchandise information base provided by an embodiment of the present invention;
FIG. 3 is an example of calculating a collocation degree of goods according to an embodiment of the present invention;
FIG. 4 is an example of a list of recommended collocation merchandise provided by an embodiment of the invention;
FIG. 5 is an example of a recommended merchandise collocation combination according to an embodiment of the present invention;
FIG. 6 is another flowchart of a method for recommending matching of an object according to an embodiment of the present invention;
FIG. 7 is a flowchart of generating a user collocation preference table according to an embodiment of the present invention;
fig. 8(a) and 8(b) are examples of generating a candidate product matching pair and a candidate product feature matching pair respectively according to an embodiment of the present invention;
FIG. 9 is an example of candidate collocation filtering provided by embodiments of the present invention;
FIG. 10 is an example of a user preference collocation collection provided by an embodiment of the invention;
fig. 11(a) and 11(b) show examples of calculating preference values of a user for a pairing when the pairing is a commodity pairing and a commodity feature pairing, respectively;
FIG. 12 is an example of predicting user preferences for non-favorite pairings provided by an embodiment of the invention;
FIG. 13 is a general block diagram of a product matching recommendation system according to an embodiment of the present invention;
fig. 14 is a schematic structural diagram of an apparatus for recommending matching of an object according to an embodiment of the present invention;
fig. 15 is another schematic structural diagram of a collocation recommending apparatus for an object according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The embodiment of the invention provides a method for recommending matching of objects, aiming at realizing matching recommendation of commodities in scenes such as network commodity recommendation and the like, so that the commodity recommendation function of a network mall is enriched, and the user experience is improved. Referring to fig. 1, a flow chart of a collocation recommendation method for an object according to the present invention may include the following steps:
step 101, obtaining a current object to be collocated of a target user.
A typical scene aimed by the scheme of the invention is a network commodity recommendation scene during network shopping. The embodiment of the present invention will be described in detail with this exemplary scenario as an example.
In this scenario, the target user may be a user currently performing commodity browsing or commodity selection in the network mall, and for the part of users, the present invention may recommend other commodities matched with the browsed or selected commodity to the user in real time through a scheme provided next, for example, when the user browses a certain jacket, trousers and shoes matched with the jacket well in terms of color and style may be recommended.
Correspondingly, the current object to be collocated may be a commodity currently browsed or selected by the target user.
102, determining at least one candidate object having an association relation with the current object based on a user collocation preference table which is formulated and stored in advance; the user collocation preference table comprises preference value information of a plurality of users on a plurality of collocation pairs, the collocation pairs are object collocation pairs consisting of two objects or feature collocation pairs consisting of two object features, and the candidate objects and the current object or the candidate objects and the corresponding features of the current object are in mutual collocation pairs in the user collocation preference table.
In order to implement the object collocation recommendation, such as the collocation recommendation of network commodities, the invention pre-establishes and stores a user collocation preference table, which includes preference value information of each user for all possible collocation pairs, the collocation pairs can be object collocation pairs or feature collocation pairs (collocation pairs of object features), taking commodities as an example, the collocation pairs can be collocation pairs composed of two different categories of commodities, such as { jack 2, troussers 1}, { jack 2, bag3} and the like, or collocation pairs composed of corresponding features of two different categories of commodities, the commodity features can be colors, styles or materials of certain categories of commodities, correspondingly, the feature collocation pairs can be collocation pairs composed of features in the aspects of colors, styles or materials and the like of different categories of commodities, such as { jacket-pink, pants-black }, { dress-casual, bag-linen }, and so on.
All possible matching pairs (object matching pairs or feature matching pairs) in the user matching preference table can be obtained based on an object information base such as a commodity information base of a network mall. The preference value information of the user to each matching pair can be obtained by analyzing the user behavior, such as the collection behavior of the user, and meanwhile, for a part of matching pairs which cannot obtain corresponding preference values by analyzing the user behavior, the preference value information of the user to the part of matching pairs can be predicted by combining with a corresponding algorithm. This section will be explained in detail in the following examples.
The preference value information of the user for the matching pair reflects the preference degree of the user for the matching pair, and in practical application, the preference value information may be a specific calculated numerical value or one of a plurality of predetermined preference levels, and the like, which is not limited by the present invention. Preferably, in this embodiment, the preference value information is a calculated specific numerical value, where the larger the value of the preference value of a certain matching pair by the user is, the higher the preference degree of the matching pair by the user is.
The commodity information base is used for storing detailed information of all commodities, including names, categories, various characteristics (such as colors, styles and materials), suitable genders and the like of the commodities, and referring to fig. 2, fig. 2 shows a specific example of the commodity information base, in which the commodity information base of a certain merchant includes various characteristics of clothing commodities. For example, "jack 1" is the name of a specific article, belonging to the category of jacket, the color is black, the style is leisure style, and is suitable for women.
On the basis of pre-establishing and storing a user collocation preference table, the step can determine each candidate object having an incidence relation with the current object to be collocated through the following processing procedures:
1) matching pairs in the user matching preference table are the conditions of object matching pairs:
finding out at least one candidate object matching pair from a user matching preference table, wherein the candidate object matching pair comprises the current object; determining other objects except the current object from the at least one candidate object collocation pair as candidate objects;
2) the matching in the user matching preference table is the condition of characteristic matching:
finding out at least one candidate feature matching pair from a user matching preference table, wherein the candidate feature matching pair comprises the corresponding feature of the current object; and determining other features except the features of the current object from the at least one candidate feature matching pair, and determining an object meeting at least one feature of the other features as a candidate object.
Specifically, taking the product collocation as an example, if the collocation pair in the user collocation preference table is a product collocation pair, and it is assumed that each row in the table represents one user, each column represents one product collocation pair, and the unit at the intersection of the rows and the columns records the preference value information of the product collocation pair corresponding to the user pair in the corresponding row, each column containing the current product to be collocated can be found from all the columns of the user collocation preference table, and finally, other products except the current product to be collocated are extracted from the obtained column collocation pairs to serve as candidate products.
If the matching pair in the user matching preference table is a feature matching pair, and each row in the table is assumed to represent one user, each column represents one feature matching pair, the unit at the intersection of the rows and the columns records the preference value information of the corresponding row of the user to the corresponding column of the feature matching pair, the matching feature of the current commodity (such as a jacket) to be matched can be obtained by searching the commodity information base, each column (namely the feature matching pair) containing the current commodity matching feature is found from all the columns of the user matching preference table, the other commodity matching feature except the current commodity feature in each obtained column matching pair is taken out, and finally each commodity meeting the commodity matching feature is determined to be a candidate commodity by searching the commodity information base.
Here, it should be noted that, in practical applications, generally, a preference value representing a corresponding preference degree may be matched for each possible collocation pair by analyzing user behaviors and calculating the preference value, or predicting the preference value, but a case that the preference value cannot be predicted may still exist theoretically is not excluded, and for this case, a part of the collocation pairs for which the preference value cannot be predicted is treated with 0 (or null or other special characters, etc.) at a corresponding preference value position, so as to represent that the user has no preference or unknown preference for the collocation pair.
And 103, screening at least one target object most matched with the current object from the at least one candidate object by utilizing a preset screening rule based on the corresponding preference value information of the matching pair where the candidate object and the current object are located in the user matching preference table.
In an actual application scenario, for example, in a network commodity recommendation scenario, the determined candidate objects are often more, and in view of this, the present embodiment proposes to adopt a predetermined screening rule to screen out at least one target object collocated with the current object from each candidate object instead of all candidate objects for recommendation.
In order to realize the screening of the candidate objects, the present embodiment proposes a concept of collocation degree, the collocation degree of the two objects reflects the degree to which the two objects can be collocated with each other, and the present embodiment assumes that the larger the collocation degree value of the two objects is, the higher the degree to which the two objects can be collocated with each other is.
On the basis, the present embodiment provides the following three alternative ways of obtaining collocation degree (one of the following three methods may be selected) by taking the commercial product as an example:
1) the individual matching degree:
the principle is as follows: different users may have different knowledge or preferences as to whether or not two goods can be matched or how well they can be matched. In view of this, the present embodiment provides the following method to obtain the personality collocation degree of two commodities X1 and X2 for a certain user U:
if the collocation pair in the user collocation preference table is a commodity collocation pair, searching a unit with a behavior U and a column of { X1, X2} or { X2, X1} from the user collocation preference table, and taking out a preference value of the unit as the personalized collocation degree of two commodities X1 and X2 to a certain user U;
if the collocation pair in the user collocation preference table is a commodity feature collocation pair, the collocation features F1 and F2 of two commodities X1 and X2 are obtained according to a commodity information base, then a unit with a row U and a column { F1, F2} or { F2, F1} is searched in the user collocation preference table, and a preference value is taken out to serve as the individual collocation degree of the two commodities X1 and X2 to a certain user U.
2) The common matching degree:
the principle is as follows: most users consider it to be applicable to all users if they consider two goods to be able to be matched.
Specifically, if the collocation pair in the user collocation preference table is a commodity collocation pair, units listed as { X1, X2} or { X2, X1} in all rows are searched from the user collocation preference table, and preference values corresponding to the units are summed to serve as the common collocation degree of two commodities X1 and X2;
if the collocation pair in the user collocation preference table is a commodity feature collocation pair, the collocation features F1 and F2 of two commodities X1 and X2 are obtained according to a commodity information base, then units listed as { F1, F2} or { F2, F1} in all rows are found from the user collocation preference table, and the preference values corresponding to the units are summed to serve as the common collocation degree of the two commodities X1 and X2.
3) Combining the individual collocation degree with the common collocation degree:
combination mode 1: when the result is not obtained according to the individual collocation degree (if the corresponding preference value is unknown), calculating the result by using the common collocation degree;
combination mode 2: and carrying out weighted summation on the individual collocation degree and the common collocation degree.
That is, the collocation degree is a individual collocation degree calculation result + b common collocation degree calculation result, where a and b are weighting factors of the individual collocation degree and the common collocation degree, respectively, and can be set manually.
Fig. 3 shows an example of the calculation of the degree of matching of goods. The category-color features of commercial trosers 1 and bag2 are "trouser-black" and "bag-black", respectively, and in the User collocation preference table, for the feature pairing { trouser-black, bag-black }, the preference value of User1 is 3, and the preference value of User2 is 2, so the personality collocation degree of trosers 1 and bag2 to User1 is 3, and the common collocation degree of trosers 1 and bag2 is 5 (assuming that two users of User1 and User2 are shared).
On the basis of calculating the matching degree of each candidate object and the current object, a predetermined number of candidate objects (which can be set manually) with the highest matching degree with the current object can be selected from the candidate objects as a final target object.
For example, taking the product as an example, the candidate products may be sorted in descending order according to the matching degree of the candidate products with the current product, and then m (a predetermined number that can be set manually) candidate products may be sequentially taken out from the sorted sequence as the target product to be recommended finally.
And 104, recommending the at least one target object to a target user in a preset display mode.
The predetermined mode may be an object list mode or an object matching and combining mode, and the like, which is not limited in this embodiment.
And if the target objects obtained by screening are recommended to the target user in an object list mode, arranging the target objects obtained by screening at the corresponding positions of the object list in a descending order according to the collocation degree of the target objects with the current object for recommendation.
Referring to fig. 4, fig. 4 shows an example of a recommended collocated item list. The User2 selects the commercial troussers 1, firstly obtains candidate commercial jackets 2, bag2, bag1 and jackets 1 from a User collocation preference table, then respectively calculates the collocation degree of each candidate commercial with troussers 1, referring to the User collocation preference table in fig. 4, it can be known that the individual collocation degree values corresponding to the candidate commercial are 4, 2, 3 and 3 respectively, and finally a recommended commercial list is obtained by descending and ordering according to the collocation degree: jack 2, bag1, jack 1, bag2, and recommend this list to User 2. Wherein, in this example, it is assumed that the predetermined number m is not less than 4.
If the target objects obtained by screening are recommended to the target user in an object matching combination mode, the categories of the current commodity and the m target commodities to be recommended can be obtained based on the commodity information base, then one target commodity is selected from each category and matched with the current commodity, and commodities of different categories can form a commodity matching combination. Then, each commodity collocation combination is verified, the collocation degree of any two commodities contained in the commodity collocation combination is obtained by utilizing the commodity collocation degree obtaining method provided above, and if the collocation degree is smaller than a certain set threshold value, the commodity collocation combination is deleted. And finally recommending the rest commodity collocation combination as a final recommendation result to the target user.
Referring to fig. 5, fig. 5 shows an example of a recommended goods collocation combination. The User2 selects the commodity troussers 1, firstly obtains target commodity lists to be recommended, namely socket 2, bag1, socket 1 and bag2, and then the target commodity lists are grouped according to categories to obtain 4 commodity collocation combinations: { jacket2, troussers 1, bag1}, { jacket2, troussers 1, bag2}, { jacket1, troussers 1, bag1}, { jacket1, troussers 1, and bag2}, on the basis of which verification is performed, assuming that the set collocation degree threshold is 3, since the troussers 1 and bag2 have a collocation degree of 2 with respect to the User2, { jacket2, troussers 1, bag2} and { jack 1, troussers 1, and bag2}, thereby finally combining the commercial collocation { jacket2, troussers 1, bag1}, { jacket1, troussers 1, and bag1} to the target User.
Therefore, the embodiment realizes that other objects which can be used in a matching way with the current object of the target user are recommended to the target user in a corresponding display way, such as an object list way or an object matching and combining way.
According to the object collocation recommending method provided by the embodiment of the invention, after the current object to be collocated of the target user is obtained, at least one candidate object having an incidence relation with the current object is determined based on a user collocation preference table which is formulated and stored in advance, at least one target object collocated with the current object is screened from the at least one candidate object by using a preset screening rule, and finally, the at least one target object collocated with the current object is recommended to the target user in a preset display mode. Therefore, the invention realizes a scheme capable of matching and recommending the object, and can realize matching and recommending of the network commodities by utilizing the scheme of the invention aiming at scenes such as network commodity recommendation and the like, thereby effectively enriching the commodity recommendation function of the network shopping mall and improving the user experience.
In another embodiment of the present invention, referring to the flowchart of an object matching recommendation method shown in fig. 6, the method of the present invention may further include a preprocessing step 101' of generating a user matching preference table, that is:
step 101': and generating a user collocation preference table.
Next, a generation process of the user collocation preference table will be explained in detail. Referring to fig. 7, the process of generating the user collocation preference table may include:
step 701, determining each possible object matching pair or feature matching pair.
Firstly, all candidate object matching pairs or feature matching pairs can be generated according to the object information base, so that a candidate object matching pair complete set or a candidate feature matching pair complete set is formed. It should be noted that the candidate object matching pair or the candidate feature matching pair in this embodiment is different from the candidate object matching pair or the candidate feature matching pair in the previous embodiment, and the candidate object matching pair or the candidate feature matching pair in this embodiment is specifically a matching pair obtained by pairwise combining objects or object features according to the object information base.
Taking the commodity as an example, if the collocation pair is defined as a commodity collocation pair, then two combinations of all commodities in the commodity information base can be performed: if N commodities are provided, the total number of N x (N-1)/2 candidate commodity matching pairs is obtained.
If the matching pair is defined as the commodity feature, the repeated records of the matching feature values can be removed from the commodity information base, and then the matching feature values of the rest records are combined pairwise to obtain each candidate feature matching pair.
Fig. 8(a) and 8(b) show examples of candidate matching pair generation. Assuming that there are 5 commodities in the commodity information base and the collocation pair in fig. 8(a) is defined as a commodity collocation pair, 10 candidate collocation pairs are generated by pairwise combination. The matching pair in fig. 8(b) is defined as a commodity feature matching pair (category-color), and repetition is first removed, and since the matching feature values of two commodities in the commodity information base are "package-red", every two of the remaining 4 commodities are combined after the repetition is removed, and 6 candidate feature matching pairs are obtained.
On this basis, all candidate pairings are filtered, and those impossible pairings are deleted. The method filters all collocation pairs based on commodity characteristic value matching, and comprises the following specific steps:
1) filtering according to the matching characteristic of the matched commodity types.
The same type of goods cannot be matched, for example, a certain jacket cannot be matched with another jacket, and the different types of goods but different types of big items cannot be matched, for example, a certain jacket cannot be matched with a certain bed sheet, because the jacket belongs to the big type of clothes, and the bed sheet belongs to the big type of bedding, so that each matched pair is filtered.
Fig. 9 shows an example of candidate collocation filtering, and for a candidate collocation pair "{ jack 1, jack 2 }", since the commodity categories of jack 1 and jack 2 are both jackets, they cannot be collocated and are filtered.
2) And filtering according to the characteristics of consistent characteristics such as style, suitable gender and the like of the matched pair.
The step is optional, and in practical application, whether the step is adopted for filtering or not can be considered according to the requirement of filtering granularity.
By filtering the various impossible pairings, the respective possible pairings (object pairings or feature pairings) can be obtained.
The next steps are to calculate the preference value of each user (such as each user of the network mall) to each possible matching pair, so as to provide the preference information basis of the user for the matching recommendation of the goods.
Step 702, obtaining the collection information of each user for the object matching combination, and calculating the preference value of each user for the object matching pair or the feature matching pair correspondingly contained in the collected object matching combination based on the collection information of each user for the object matching combination.
In the online shopping scene, the user often collects the interested commodity information with corresponding preference so as to facilitate subsequent review or purchase.
In practical applications, tools may be developed to allow a user to add one or more product matching combinations to the favorite of the user, that is, allow the user to collect product matching combinations (different from the prior art that only allows the user to collect single product information), where each product matching combination includes a set (at least two) of products that can be matched with each other.
For example, for clothing goods, a personalized wardrobe tool can be developed, which creates a virtual portrait, and the user can select a certain piece of clothing from the goods area to put on the virtual portrait or take off the certain piece of clothing from the virtual portrait, so that individual goods matching combinations can be generated in an intuitive manner. Optionally, in other embodiments of the present invention, the user may also be enabled to browse the favorites of other users and select a favorite collocation combination from the favorites to join the favorite. Alternatively, the user may also score each product mix of the collection (e.g., from one star to five stars, or from 1 to 10 points, etc.).
Referring to FIG. 10, FIG. 10 shows an example of a User preference collocation collection in which a User "User 1" has collected two product collocation combinations, the first product collocation combination being { jack 1, troussrs 1, bag1}, and the second product collocation combination being { jack 2, troussrs 1, bag2 }.
On the basis that the user can collect the commodity collocation combination, the favorite of each user is analyzed to obtain the collection information of the user on the commodity collocation combination, and the preference value of each user on the collocation pair correspondingly contained in the collected commodity collocation combination is calculated on the basis of the collection information of each user on the commodity collocation combination, wherein each collocation pair contains two commodities or commodity characteristics capable of being collocated.
The collection information may include, but is not limited to: the commodity information and combination relation correspondingly contained in the commodity collocation combination collected by the user, the user collection times, the user browsing times and/or the scores of the user on the collected commodity collocation combination, and the like.
If the matching pair is a commodity matching pair, the preference value of the user to the commodity matching pair can be calculated through the following processing procedures:
1) and extracting the matching pairs of the collected commodities in the matching combination of the collected commodities of the user.
Analyzing each commodity matching combination collected by each user, extracting all the commodity combinations from each commodity matching combination, then removing the repeated commodity combinations, and finally obtaining each collected commodity matching pair of the user.
2) And calculating the preference value of the user to the matching pair of the collected commodities.
The preference value of each user for each collected commodity matching pair can be calculated based on the collection information such as the collection times, the browsing times and the scores of the users. The calculation formula of the preference value of a certain user to the collected commodity matching pair P is as follows:
in the formula, G1，G2，…，GmShowing the collocation combination of each commodity collected by the user,this embodiment shows that only the commodity collocation combination, browse Times (G), including the collected commodity collocation pair P is analyzedi) RatingScore (G) for browsing the ith product collocation combination by the useri) And the MaxRatingScore is the full score value of the user score for the user to score the ith commodity collocation combination.
According to the formula, the higher the collection times, browsing times and scores of the user on the matched pairs are, the higher the preference value is.
3) And (4) classifying preference values.
The preference values may be divided in size into several preference levels, each representing a different degree of preference, respectively. Such as dividing preference values into common levels 1-5, etc. This step is an optional step.
If the matching pair is a commodity feature matching pair, the preference value of the user to the commodity feature matching pair can be calculated through the following processing procedures:
1) and extracting the matching pairs of the collected commodity characteristics in the matching combination of the collected commodities of the user.
Analyzing each commodity matching combination collected by each user, and extracting the value of the characteristic corresponding to each commodity from each commodity matching combination based on the commodity information base, so as to obtain each commodity characteristic matching combination collected by each user. And then, extracting all pairwise characteristic combinations from each commodity characteristic matching combination, removing repeated pairwise characteristic combinations, and finally obtaining each collected commodity characteristic matching pair of the user.
2) And calculating the preference value of the matching pair of the characteristics of the collected commodities by the user.
Specifically, the preference value of each user for each matching pair of the collected commodity features can be calculated based on the collection information such as the collection times, the browsing times and the scores of the matching combinations of the collected commodities of the user. The calculation formula of the preference value of a certain user to the collected commodity feature matching pair F is as follows:
in the formula, G'iCommodity matching combination G for user collectioniThe corresponding commodity characteristics are matched and combined,it is shown that only the matching combination of the commodity features, browse Times (G), including the matching pair of the collected commodity features F is analyzed in this embodimenti) RatingScore (G) for browsing the ith product collocation combination by the useri) And the MaxRatingScore is the full score value of the user score for the user to score the ith commodity collocation combination.
3) And (4) classifying preference values.
The preference values may be divided in size into several preference levels, each representing a different degree of preference, respectively. Such as dividing preference values into common levels 1-5, etc. This step is an optional step.
Referring to fig. 11(a) and 11(b), fig. 11(a) and 11(b) show examples of calculating a preference value of a user for a pairing when the pairing is a product pairing and a product feature pairing, respectively.
In fig. 11(a), the matching pairs are commodity matching pairs, and the User "User 2" has collected two commodity matching combinations: { jack 2, troussers 1} and { jack 2, troussers 1, bag2}, where the first combination has a browsing count of 2 and a score of 10, and the second combination has a browsing count of 3 and a score of 8 (10 points full). According to the method for calculating the preference value, firstly, the collected commodity matching pairs are extracted to obtain 3 collected commodity matching pairs: { jacks 2, troussers 1}, { jacks 2, bag2} and { troussers 1, bag2}, and then preference calculations were performed, { jacks 2, troussers 1} occurred in the two commodity collocation combinations collected by the user, with preference values of (2/3) × (10/10) + (3/3) × (8/10) ═ 1.47, { jacket2, bag2} and { troussers 1, bag2} occurring only in the second combination, with preference values of (3/3) × (8/10) × 0.8. The preference values are finally sorted into 5 preference classes.
In fig. 11(b), the matching pairs are commodity feature matching pairs (specifically, pairwise matching pairs of commodity features such as "category-color"), and for the favorite of the User "User 2", the category and color values of each commodity to be collected are first taken out based on the commodity information library, so as to obtain 2 commodity matching feature combinations: { jacket-pink, trousers-black } and { jacket-pink, trousers-black, bag-black }, on this basis, 3 collected commodity feature pairings can be obtained: { jacket-pink, trousers-black }, { jacket-pink, bag-black } and { trousers-black, bag-black }, and preference values were then calculated, wherein { jacket-pink, trousers-black } occurred in two commodity collocation feature combinations, with preference values of 1.47, { jacket-pink, bag-black } and { trousers-black, bag-black } occurring only in the second commodity collocation feature combination, with preference values of 0.8. The preference values are finally sorted into 5 preference classes.
And 703, predicting the preference value of the matching pair of which the preference value is not calculated in each possible object matching pair or feature matching pair by each user by using a preset algorithm.
According to the preference value calculation method provided in step 702, the preference value of the user to the favorite pairings (the pairings correspondingly included in the commodity collocation combination collected by the user) can be calculated, and for the pairings not collected by the user in the user collocation preference table (the pairings not included in the commodity collocation combination collected by the user), the preference value of the user to the partial favorite pairings is predicted by adopting a corresponding prediction method.
In the embodiment, the collaborative filtering algorithm is utilized, the preference of each user for all non-favorite pairs is predicted based on the preference of the user for the favorite pairs, and finally, the preference value of each user for each favorite pair is stored in the user collocation preference table.
The embodiment specifically provides the following method to predict the preference values of the user for all the non-favorite pairings:
1) and initializing a user collocation preference table.
Combining the calculated preference values of all the user pairs for the favorite matching pairs together, and adding other possible matching pairs which are not included into a new column in the user matching preference table, thereby generating an m × n initial user matching preference table, wherein m is the number of users, n is the number of matching pairs, each row in the table represents one user, each column represents one matching pair, the unit at the intersection of each row and column records the preference value of a certain user to a certain matching pair, and the preference value is unknown to be null.
2) And (4) predicting unknown preference.
A collaborative filtering algorithm is used to calculate individual unknown preference values. The theory of the collaborative filtering algorithm is as follows: and comparing the similarity of the collection preferences of the target user and other users to identify a group of users with similar preferences, and if the users with similar preferences like a collocation with the target user, considering that the target user also likes the collocation so as to predict the preference value of the target user for the collocation.
Referring to FIG. 12, FIG. 12 illustrates an example of a user's preference prediction for other non-favorite pairs. Although the collocation pair { jack 2 and bag2} is not collected by the User1, the preference of the User1 to { jack 2 and bag2} can be predicted by calculating the collection similarity with other users.
Step 704, generating a user matching preference table according to the possible object matching pairs or feature matching pairs and the preference values of the possible object matching pairs or feature matching pairs of the users.
On the basis of calculating the preference value of each user for the collected matching pair and predicting the preference value of each user for the non-collected matching pair, the obtained preference values can be filled to the corresponding positions of the user matching preference table, so that the user matching preference table capable of representing the preference information of each user for each matching pair (commodity matching pair or commodity characteristic matching pair) is generated, and a basis can be provided for subsequent commodity matching recommendation to the user.
Based on the generation process of the user collocation preference table provided in this embodiment, a general block diagram for realizing collocation recommendation of a commodity according to the present invention can be referred to as shown in fig. 13.
The invention provides a collocation recommending device for an object in the next another embodiment, aiming at realizing collocation recommendation of commodities in scenes such as network commodity recommendation and the like, thereby enriching the commodity recommending function of a network mall and improving the user experience. Referring to fig. 14, a schematic structural diagram of an object collocation recommendation device according to the present invention may include:
the device comprises an acquisition unit 1, a matching unit and a matching unit, wherein the acquisition unit is used for acquiring a current object to be matched of a target user; the determining unit 2 is configured to determine at least one candidate object having an association relationship with the current object based on a user collocation preference table which is formulated and stored in advance; the user collocation preference table comprises preference value information of a plurality of users on a plurality of collocation pairs, the collocation pairs are object collocation pairs consisting of two different types of objects or feature collocation pairs consisting of features of the two different types of objects, and the candidate objects and the current object or the candidate objects and the corresponding features of the current object are in collocation pairs in the user collocation preference table; a screening unit 3, configured to screen, based on preference value information corresponding to a matching pair where a candidate object and a current object are located in the user matching preference table, at least one target object most matched with the current object from the at least one candidate object by using a predetermined screening rule; and the collocation recommending unit 4 is used for recommending the at least one target object to the target user in a preset display mode.
In an implementation manner of the embodiment of the present invention, the determining unit is further configured to:
if the collocation pair in the user collocation preference table is an object collocation pair, then: finding out at least one candidate object matching pair from a user matching preference table, wherein the candidate object matching pair comprises the current object; determining other objects except the current object from the at least one candidate object collocation pair as candidate objects; if the matching pair in the user matching preference table is a feature matching pair, then: finding out at least one candidate feature matching pair from a user matching preference table, wherein the candidate feature matching pair comprises the corresponding feature of the current object; and determining other features except the features of the current object from the at least one candidate feature matching pair, and determining an object meeting at least one feature of the other features as a candidate object.
In an implementation manner of the embodiment of the present invention, the screening unit is further configured to: finding out preference value information of each candidate object and the object matching pair where the current object is located or the feature matching pair corresponding to the target user from the user matching preference table, and/or preference value information corresponding to all users; calculating the collocation degree of each candidate object and the current object based on the obtained preference value information; and screening out a preset number of candidate objects with the highest matching degree with the current object as target objects.
In an implementation manner of the embodiment of the present invention, the collocation recommending unit is further configured to: recommending the at least one target object to a target user in an object list mode; or recommending the at least one target object to a target user in an object collocation combination mode, wherein each recommended object collocation combination is a collocation combination of the at least one target object and the current object.
In an implementation manner of the embodiment of the present invention, referring to fig. 15, the apparatus may further include: and the preprocessing unit 5 is used for generating a user collocation preference table.
The pre-processing unit generating a user collocation preference table comprises: determining each possible object or feature pairing; acquiring collection information of each user on the object matching combination, and calculating preference values of each user on object matching pairs or feature matching pairs correspondingly contained in the collected object matching combinations based on the collection information of each user on the object matching combination; predicting the preference value of the matched pair of which the preference value is not calculated in each possible object matched pair or feature matched pair by each user by using a preset algorithm; and generating a user matching preference table according to the possible object matching pairs or the feature matching pairs and the preference values of the users to the possible object matching pairs or the feature matching pairs.
It should be noted that, the description of the object collocation recommending apparatus according to the embodiment is similar to the description of the method above, and the beneficial effects of the method are described, for the technical details of the object collocation recommending apparatus of the present invention that are not disclosed in the embodiment, please refer to the description of the method embodiment of the present invention, which is not repeated herein.
In summary, the object matching recommendation scheme of the present invention has the following advantages: the invention provides a commodity collocation recommending method based on machine learning, which can be used for recommending other commodities which are matched with commodities browsed by a user or selected commodities individually, can improve user experience, enriches the commodity recommending function of a network mall, can be more suitable for user requirements in collocation type recommendation of the commodities, and is more easy to attract users to purchase compared with general commodity recommendation.
It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other.
For convenience of description, the above system or apparatus is described as being divided into various modules or units by function, respectively. Of course, the functionality of the units may be implemented in one or more software and/or hardware when implementing the present application.
From the above description of the embodiments, it is clear to those skilled in the art that the present application can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present application may be essentially or partially implemented in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments or some parts of the embodiments of the present application.
Finally, it is further noted that, herein, relational terms such as first, second, third, fourth, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.
Claims (10)
1. A collocation recommendation method for an object is characterized by comprising the following steps:
obtaining a current object to be collocated of a target user;
determining at least one candidate object having an association relation with the current object based on a user collocation preference table which is formulated and stored in advance; the user collocation preference table comprises preference value information of a plurality of users on a plurality of collocation pairs, the collocation pairs are object collocation pairs consisting of two different types of objects or feature collocation pairs consisting of features of the two different types of objects, and the candidate objects and the current object or the candidate objects and the corresponding features of the current object are in collocation pairs in the user collocation preference table;
selecting at least one target object which is most matched with the current object from the at least one candidate object by utilizing a preset selection rule based on the preference value information corresponding to the matching pair where the candidate object and the current object are located in the user matching preference table;
and recommending the at least one target object to the target user in a preset presentation mode.
2. The method of claim 1, wherein the determining at least one candidate object having an association relationship with the current object based on a pre-established and stored user collocation preference table comprises:
if the collocation pair in the user collocation preference table is an object collocation pair, then:
finding out at least one candidate object matching pair from a user matching preference table, wherein the candidate object matching pair comprises the current object;
determining other objects except the current object from the at least one candidate object collocation pair as candidate objects;
if the matching pair in the user matching preference table is a feature matching pair, then:
finding out at least one candidate feature matching pair from a user matching preference table, wherein the candidate feature matching pair comprises the corresponding feature of the current object;
and determining other features except the features of the current object from the at least one candidate feature matching pair, and determining an object meeting at least one feature of the other features as a candidate object.
3. The method of claim 1, wherein the selecting at least one target object most collocated with the current object from the at least one candidate object based on the corresponding preference value information of the collocation pair of the candidate object and the current object in the user collocation preference table and by using a predetermined selection rule comprises:
finding out preference value information of each candidate object and the object matching pair where the current object is located or the feature matching pair corresponding to the target user from the user matching preference table, and/or preference value information corresponding to all users;
calculating the collocation degree of each candidate object and the current object based on the obtained preference value information;
and screening out a preset number of candidate objects with the highest matching degree with the current object as target objects.
4. The method according to claim 1, wherein the recommending the at least one target object to the user in a predetermined presentation manner comprises:
recommending the at least one target object to a target user in an object list mode; or,
recommending the at least one target object to a target user in an object collocation combination mode, wherein each recommended object collocation combination is a collocation combination of the at least one target object and the current object.
5. The method according to any one of claims 1-4, further comprising, before the obtaining of the current object to be collocated of the target user, a preprocessing step of: generating a user collocation preference table;
the generating of the user collocation preference table comprises the following steps:
determining each possible object or feature pairing;
acquiring collection information of each user on the object matching combination, and calculating preference values of each user on object matching pairs or feature matching pairs correspondingly contained in the collected object matching combinations based on the collection information of each user on the object matching combination;
predicting the preference value of the matched pair of which the preference value is not calculated in each possible object matched pair or feature matched pair by each user by using a preset algorithm;
and generating a user matching preference table according to the possible object matching pairs or the feature matching pairs and the preference values of the users to the possible object matching pairs or the feature matching pairs.
6. An arrangement for recommending a collocation of objects, comprising:
the device comprises an acquisition unit, a matching unit and a matching unit, wherein the acquisition unit is used for acquiring a current object to be matched of a target user;
the determining unit is used for determining at least one candidate object which has an incidence relation with the current object based on a user collocation preference table which is formulated and stored in advance; the user collocation preference table comprises preference value information of a plurality of users on a plurality of collocation pairs, the collocation pairs are object collocation pairs consisting of two different types of objects or feature collocation pairs consisting of features of the two different types of objects, and the candidate objects and the current object or the candidate objects and the corresponding features of the current object are in collocation pairs in the user collocation preference table;
a screening unit, configured to screen, based on preference value information corresponding to a matching pair where a candidate object and a current object are located in the user matching preference table, at least one target object most matched with the current object from the at least one candidate object by using a predetermined screening rule;
and the collocation recommending unit is used for recommending the at least one target object to the target user in a preset display mode.
7. The apparatus of claim 6, wherein the determining unit is further configured to:
if the collocation pair in the user collocation preference table is an object collocation pair, then:
finding out at least one candidate object matching pair from a user matching preference table, wherein the candidate object matching pair comprises the current object; determining other objects except the current object from the at least one candidate object collocation pair as candidate objects;
if the matching pair in the user matching preference table is a feature matching pair, then:
finding out at least one candidate feature matching pair from a user matching preference table, wherein the candidate feature matching pair comprises the corresponding feature of the current object; and determining other features except the features of the current object from the at least one candidate feature matching pair, and determining an object meeting at least one feature of the other features as a candidate object.
8. The apparatus of claim 6, wherein the screening unit is further configured to:
finding out preference value information of each candidate object and the object matching pair where the current object is located or the feature matching pair corresponding to the target user from the user matching preference table, and/or preference value information corresponding to all users; calculating the collocation degree of each candidate object and the current object based on the obtained preference value information; and screening out a preset number of candidate objects with the highest matching degree with the current object as target objects.
9. The apparatus of claim 6, wherein the collocation recommendation unit is further configured to:
recommending the at least one target object to a target user in an object list mode; or recommending the at least one target object to a target user in an object collocation combination mode, wherein each recommended object collocation combination is a collocation combination of the at least one target object and the current object.
10. The apparatus of any one of claims 6-9, further comprising:
the preprocessing unit is used for generating a user collocation preference table;
the preprocessing unit generates a user collocation preference table, comprising:
determining each possible object or feature pairing; acquiring collection information of each user on the object matching combination, and calculating preference values of each user on object matching pairs or feature matching pairs correspondingly contained in the collected object matching combinations based on the collection information of each user on the object matching combination; predicting the preference value of the matched pair of which the preference value is not calculated in each possible object matched pair or feature matched pair by each user by using a preset algorithm; and generating a user matching preference table according to the possible object matching pairs or the feature matching pairs and the preference values of the users to the possible object matching pairs or the feature matching pairs.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710465115.1A CN107256513A (en) | 2017-06-19 | 2017-06-19 | Method and device is recommended in a kind of collocation of object |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710465115.1A CN107256513A (en) | 2017-06-19 | 2017-06-19 | Method and device is recommended in a kind of collocation of object |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107256513A true CN107256513A (en) | 2017-10-17 |
Family
ID=60023277
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710465115.1A Pending CN107256513A (en) | 2017-06-19 | 2017-06-19 | Method and device is recommended in a kind of collocation of object |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107256513A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108022155A (en) * | 2017-12-07 | 2018-05-11 | 北京小米移动软件有限公司 | Jewel accessory recommends method and device |
| CN108198051A (en) * | 2018-03-01 | 2018-06-22 | 口碑(上海)信息技术有限公司 | Across the Method of Commodity Recommendation and device of merchandise classification |
| CN108596705A (en) * | 2018-03-23 | 2018-09-28 | 郑州大学西亚斯国际学院 | A kind of commodity suitable for e-commerce recommend method and system with information classification |
| WO2019174549A1 (en) * | 2018-03-12 | 2019-09-19 | 北京京东尚科信息技术有限公司 | Information recommendation method and apparatus |
| CN110415063A (en) * | 2018-07-31 | 2019-11-05 | 北京京东尚科信息技术有限公司 | Method of Commodity Recommendation, device, electronic equipment and readable medium |
| CN110581875A (en) * | 2018-06-11 | 2019-12-17 | 阿里巴巴集团控股有限公司 | Information pushing method, device and system applied to cloud shelf |
| CN113112322A (en) * | 2021-03-11 | 2021-07-13 | 黄永辉 | Internet-based paired clothing pushing method, device, equipment and storage medium |
| CN113140278A (en) * | 2020-01-20 | 2021-07-20 | 阿里健康信息技术有限公司 | Data processing method, terminal device, server and storage medium |
| CN113742570A (en) * | 2020-12-01 | 2021-12-03 | 北京沃东天骏信息技术有限公司 | Commodity searching method, terminal equipment and server |
| CN113763114A (en) * | 2021-03-04 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Article information matching method and device and storage medium |
| CN114169952A (en) * | 2020-09-11 | 2022-03-11 | 京东方科技集团股份有限公司 | Commodity recommendation method, server, shopping cart and shopping system |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101859313A (en) * | 2009-04-08 | 2010-10-13 | 索尼公司 | Messaging device and method and program thereof |
| CN103093369A (en) * | 2011-11-03 | 2013-05-08 | 阿里巴巴集团控股有限公司 | Method and device for offering matched product based on correlation degree between products |
| CN105205684A (en) * | 2014-06-30 | 2015-12-30 | 阿里巴巴集团控股有限公司 | Recommended display method of matched products and apparatus |
| CN105469263A (en) * | 2014-09-24 | 2016-04-06 | 阿里巴巴集团控股有限公司 | Commodity recommendation method and device |
| CN106204124A (en) * | 2016-07-02 | 2016-12-07 | 向莉妮 | Personalized commercial coupling commending system and method |
| CN106294420A (en) * | 2015-05-25 | 2017-01-04 | 阿里巴巴集团控股有限公司 | The method and device of business object collocation information is provided |
| CN106547365A (en) * | 2015-09-17 | 2017-03-29 | 阿里巴巴集团控股有限公司 | The method and apparatus of commercial product recommending |
| CN106709781A (en) * | 2016-12-05 | 2017-05-24 | 姚震亚 | Personal image design and collocation purchasing device and method |
-
2017
- 2017-06-19 CN CN201710465115.1A patent/CN107256513A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101859313A (en) * | 2009-04-08 | 2010-10-13 | 索尼公司 | Messaging device and method and program thereof |
| CN103093369A (en) * | 2011-11-03 | 2013-05-08 | 阿里巴巴集团控股有限公司 | Method and device for offering matched product based on correlation degree between products |
| CN105205684A (en) * | 2014-06-30 | 2015-12-30 | 阿里巴巴集团控股有限公司 | Recommended display method of matched products and apparatus |
| CN105469263A (en) * | 2014-09-24 | 2016-04-06 | 阿里巴巴集团控股有限公司 | Commodity recommendation method and device |
| CN106294420A (en) * | 2015-05-25 | 2017-01-04 | 阿里巴巴集团控股有限公司 | The method and device of business object collocation information is provided |
| CN106547365A (en) * | 2015-09-17 | 2017-03-29 | 阿里巴巴集团控股有限公司 | The method and apparatus of commercial product recommending |
| CN106204124A (en) * | 2016-07-02 | 2016-12-07 | 向莉妮 | Personalized commercial coupling commending system and method |
| CN106709781A (en) * | 2016-12-05 | 2017-05-24 | 姚震亚 | Personal image design and collocation purchasing device and method |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108022155A (en) * | 2017-12-07 | 2018-05-11 | 北京小米移动软件有限公司 | Jewel accessory recommends method and device |
| CN108198051A (en) * | 2018-03-01 | 2018-06-22 | 口碑(上海)信息技术有限公司 | Across the Method of Commodity Recommendation and device of merchandise classification |
| US11748799B2 (en) | 2018-03-12 | 2023-09-05 | Beijing Jingdong Shangke Information Technology Co., Ltd. | Method, medium, and system for information recommendation |
| WO2019174549A1 (en) * | 2018-03-12 | 2019-09-19 | 北京京东尚科信息技术有限公司 | Information recommendation method and apparatus |
| CN110264291A (en) * | 2018-03-12 | 2019-09-20 | 北京京东尚科信息技术有限公司 | Information recommendation method and device |
| CN110264291B (en) * | 2018-03-12 | 2024-06-18 | 北京京东尚科信息技术有限公司 | Information recommendation method and device |
| CN108596705A (en) * | 2018-03-23 | 2018-09-28 | 郑州大学西亚斯国际学院 | A kind of commodity suitable for e-commerce recommend method and system with information classification |
| CN110581875A (en) * | 2018-06-11 | 2019-12-17 | 阿里巴巴集团控股有限公司 | Information pushing method, device and system applied to cloud shelf |
| CN110415063A (en) * | 2018-07-31 | 2019-11-05 | 北京京东尚科信息技术有限公司 | Method of Commodity Recommendation, device, electronic equipment and readable medium |
| CN113140278A (en) * | 2020-01-20 | 2021-07-20 | 阿里健康信息技术有限公司 | Data processing method, terminal device, server and storage medium |
| CN114169952A (en) * | 2020-09-11 | 2022-03-11 | 京东方科技集团股份有限公司 | Commodity recommendation method, server, shopping cart and shopping system |
| CN113742570A (en) * | 2020-12-01 | 2021-12-03 | 北京沃东天骏信息技术有限公司 | Commodity searching method, terminal equipment and server |
| CN113763114A (en) * | 2021-03-04 | 2021-12-07 | 北京沃东天骏信息技术有限公司 | Article information matching method and device and storage medium |
| CN113763114B (en) * | 2021-03-04 | 2024-07-16 | 北京沃东天骏信息技术有限公司 | Item information collocation method, device and storage medium |
| CN113112322A (en) * | 2021-03-11 | 2021-07-13 | 黄永辉 | Internet-based paired clothing pushing method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107256513A (en) | Method and device is recommended in a kind of collocation of object | |
| US10402917B2 (en) | Color-related social networking recommendations using affiliated colors | |
| US9183510B1 (en) | Method and system for personalized recommendation of lifestyle items | |
| KR20190117584A (en) | Method and apparatus for detecting, filtering and identifying objects in streaming video | |
| US8285715B2 (en) | System and method for the structured display of items | |
| CN109146626B (en) | Fashionable garment matching recommendation method based on user dynamic interest analysis | |
| CN107633430A (en) | A kind of Method of Commodity Recommendation based on community of colony | |
| US9727620B2 (en) | System and method for item and item set matching | |
| US20190355041A1 (en) | Cognitive fashion product recommendation system and method | |
| JP2012529122A (en) | System and method for learning user genre and style and matching products to user preferences | |
| CN106202316A (en) | Merchandise news acquisition methods based on video and device | |
| KR101825498B1 (en) | Contents recommendation system and method | |
| WO2002079942A2 (en) | System for visual preference determination and predictive product selection | |
| US11386301B2 (en) | Cluster and image-based feedback system | |
| KR102309373B1 (en) | System for building the contents automatically for planning sales and method therefor | |
| WO2020085086A1 (en) | Commodity recommendation system | |
| CN106874503B (en) | Method and device for acquiring recommended data | |
| US20090037292A1 (en) | Intelligent shopping search system | |
| CN111861678B (en) | Resource recommendation method and device, electronic equipment and storage medium | |
| Sekozawa et al. | One‐to‐one recommendation system in apparel online shopping | |
| KR20170116924A (en) | Method for recommendation service of product using relative comparison, and device and program for the same | |
| Tahir et al. | E-commerce platform based on Machine Learning Recommendation System | |
| CN109658195B (en) | Commodity display decision method | |
| KR20200057209A (en) | A system for suggesting customized books using k-means clustering and method thereof | |
| JP2016071881A (en) | Item recommendation system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171017 |